Apple specifies AI agent permission overreach and transaction completion risks in 10-K

Apple disclosed in its fiscal 2026 Form 10-K, filed with the SEC for the year ended September 26, that AI agents—both its own and third-party agents—pose business risks by exceeding their authority, disregarding instructions, and taking hard-to-reverse actions such as executing or altering code, sending communications, and completing transactions. The filing states these agents can access internal and external systems and interact with systems they lack permission to use, extending risk beyond incorrect answers to direct effects on systems and data. The disclosure matters because it formally acknowledges agentic AI as a material operational and transactional risk for a company whose ecosystem spans devices, software, and services. What to watch next is how Apple governs agent permissions and transaction finality across its platforms, and whether peer tech firms add similar language to their own filings.